英文原文.pdf

Desalination 144 (2002) 167172 Presented at the International Congress on Membranes and Membrane Processes (ICOM), Toulouse, France, July 712, 2002. 0011-9164/02/$ See front matter 2002 Elsevier Science B.V. All rights reserved *Corresponding author. Preparation and analysis of oil-in-water emulsions with a narrow droplet size distribution using Shirasu-porous-glass (SPG) membranes Goran T. Vladisavljevica,*, Helmar Schubertb aInstitute of Food Technology and Biochemistry, Faculty of Agriculture, University of Belgrade, P.O. Box 127, YU-11081 Belgrade-Zemun, Yugoslavia Tel. +381 (11) 615-315/327; Fax +381 (11) 199-711; email: gtvladisafrodita.rcub.bg.ac.yu bInstitute of Food Process Engineering, Faculty of Chemical Engineering, University of Karlsruhe (T.H.), Kaiserstrasse 12, D-76128 Karlsruhe, Germany Tel. +49 (721) 608-2497; Fax +49 (721) 694-320; email: Helmar.Schubertlvt.uni-karlsruhe.de Received 30 January 2002; accepted 13 February 2002 Abstract Shirasu-porous-glass (SPG) membranes with a mean pore size from 0.46.6 m were used to produce O/W emulsions consisting of vegetable (rape seed) oil as the dispersed phase and Span 80 dissolved in demineralized water as the continuous phase. The emulsion droplets with a mean droplet size 3.5 times larger than the mean pore size and the span of the droplet size distribution between 0.26 and 0.45 were produced using 2% emulsifier at a transmembrane pressure slightly exceeding the capillary pressure. Under these conditions the dispersed phase flux through the membrane was in the range of 0.77 lm2h1 and only about 2% of the pores were active. However, if the transmembrane pressure was considerably higher than the capillary pressure, the dispersed phase flux strongly increased and droplets with a broad droplet size distribution were produced. The hydraulic resistance of the SPG membrane was inversely proportional to the square of the mean pore size, which is in agreement with the Hagen- Poiseuille law. The membrane porosity is independent on the pore size and ranged from 5360%. Keywords: Membrane emulsification; Porous glass membrane; Oil-in-water emulsions; SPG membrane; Hydraulic membrane resistance; Membrane porosity. 168G.T. Vladisavljevic, H. Schubert / Desalination 144 (2002) 167172 1. Introduction Membrane emulsification (ME) is a new emulsification technique, especially suitable for the production of highly uniform particles or droplets of controlled mean size 1. In conventional emul- sification devices, such as high-pressure homo- genizers and rotor-stator systems, interfacial area is increased by disruption of larger droplets in a premix using high energy inputs 2. In a ME system, however, small droplets are directly formed by permeation of dispersed phase through a porous membrane into moving (flowing or stirring) continuous phase. In that way, droplet size can be more effectively controlled and the required shear stress is much smaller. The additional advantage is the possibility to obtain uniform droplets over a much wider range of mean droplet sizes. Due to uniform pores, a wide range of available mean pore sizes (0.0530 m) and the possibility of surface modification, the Shirasu-porous-glass membrane developed by Nakashima and Shimizu 3 is a potentially suitable membrane for mecha- nical emulsification. The aim of this work is to investigate the influence of pore size of the SPG membrane and some operating parameters on the emulsification result at the transmembrane pressures slightly exceeding the capillary pressure. 2. Theory The permeation of pure water through a porous membrane with a thickness of m whose pores are assumed to be capillaries of diameter dp and length lp, can be explained in terms of the Hagen- Poiseuille law 4: 2 /32 pwpwmwwtm dvlRJp= (1) where ptm is the transmembrane pressure, Rm the hydraulic membrane resistance, w the water viscosity, Jw the water flux through the membrane, and vw the mean water velocity in the pores. The substitution of vw = Jw/ and lp = m into Eq. (1) gives: )/(32)/( 2 pmwwtmm dJpR=(2) where is the mean tortuosity factor of the pores and the membrane porosity. During ME process droplets are usually formed simultaneously at only 240% of the pores 5. The fraction of active pores at any moment is given by: tmmdd pRJk=/ (3) where Jd is the dispersed phase flux through the membrane and d is the viscosity of dispersed phase. If the active pores are arranged over the membrane surface in a square array, the distance between the centers of adjacent active pores can be expressed as: 5 . 0 )/)(2/(kdz p =(4) Neglecting droplet deformation in the direction of continuous phase flow, droplets do not touch each other at the pore openings if z ddrop. Thus, the condition for unhindered droplet growth in the case of uniform pore arrangement and rigid droplets reads: 2 )/(5 . 0/ kdd pdrop (5) In order to avoid contact between two neigh- boring droplets at the pore openings, the fraction of active pores must be kept below kmax = (/4)(ddrop/ dp)0.5. The dependence of kmax on droplet/pore diameter ratio and membrane porosity is shown in Fig. 1. The porosities of SPG membranes are in the range of 0.50.6 6 and for O/W emulsions ddrop/dp ranges typically from 2.58 7. Therefore, to ensure that no coalescence can occur at the surface of the SPG membrane, the percentage of active pores must be kept below 225%, depending mostly on the droplet/pore diameter ratio. The average droplet formation time tf can be calculated as 89: G.T. Vladisavljevic, H. Schubert / Desalination 144 (2002) 167172169 )/)(/)(3/2( 3 3 , 4 2 dpf Jddkt=(6) Although the above expressions are only approximate, they are useful in an attempt to give some quantitative explanations of the experimental results. 3. Experimental The O/W emulsions have been prepared using vegetable (rape seed) oil (Floreal, Germany) with a viscosity of 58 mPas as the dispersed phase and 2% (w/w) Tween 80 (Merck) dissolved in demine- ralized water as the continuous phase. Micro- porous glass membranes were supplied from SPG Technology (Miyazaki, Japan) with a mean pore size of 0.4, 1.4, 2.5, 5.0, and 6.6 m, determined by a Shimadzu model 9320 mercury porosimeter. The SPG membrane tube (125 mm length 10 mm OD 0.7 mm wall thickness) was installed inside a laboratory made stainless steel module pro- viding the effective membrane area of 31.4 cm2. The continuous phase was recirculated in a closed loop between the membrane module and a continuous-phase reservoir using a Netzsch model 10 1 10 100 Droplet/pore diameter ratio ddrop/dp =0.1 =0.2 =0.4 =0.6 2 Fig. 1. Maximum percentage of active pores for unhindered droplet growth as a function of droplet/pore diameter ratio and membrane porosity. NL 20 Mohno-pump (Fig. 2). In the most expe- riments, the continuous phase flow rate was 305 lh1, which corresponds to a mean velocity in the membrane tube of 1.4 ms1 and tube Reynolds number of 8500. Under these conditions the shear stress at the membrane surface was 8 Pa. The oil phase was placed in the pressure vessel and was introduced at the module shell side with compressed air. The weight of oil permeated through the membrane was measured by a digital balance. The droplet size distribution was determined by a light scattering particle size analyser using PIDS technology (Coulter LS 230). 4. Results and discussion 4.1. Properties of the SPG membranes used in this study The hydraulic and morphological properties of the SPG membranes used are listed in Table 1. The hydraulic membrane resistances were calculated by measuring the pure water flux through the membrane and using Eq. (2). The membrane resistance was inversely proportional to the square of the mean pore size (Fig. 3): emulsion or continuous phase Compressed air Mohno pump Cleaning pump Balance PI PIPI m Vent Cleaning cycle Oil cleaning solution t Module V1 Fig. 2. Schematic view of the experimental set-up for crossflow membrane emulsification. The V1 valve is closed during emulsification and open during cleaning-in-place (CIP) cycle. 170G.T. Vladisavljevic, H. Schubert / Desalination 144 (2002) 167172 Table 1 The hydraulic resistances, Rm, the wall porosities, , and the mean pore tortuosity factors, for the SPG membranes used in this work Pore size, dp, m 0.41.4 2.5 5.0 6.6 Rm 109, m1 350 29 9.0 3.2 1.5 Porosity, 0.60 0.53 0.60 0.58 Tortuosity, 1.7 1.3 2.1 1.7 0.1110 1 10 100 103 Mean pore size dp / m Fig. 3. Hydraulic resistance of the SPG membrane as a function of mean pore size. 2 056. 0 = pm dR(7) where Rm and dp are in m1 and m, respectively. Eq. (7) is in accordance with Eq. (2). However, the membrane porosities measured by the pycno- metric method 10 and the mean pore tortuosity factors calculated from Eq. (2) were independent on the mean pore size (Table 1). The values listed in Table 1 are within the range of 5060%, reported earlier for a typical SPG membrane 6. As a comparison, the porosity of coating layer of composite ceramic membranes ranges typically from 3060% 5. 4.2. Preparation of O/W emulsions using the SPG membranes The experimental results obtained at ptm 1.1pcap are given in Table 2. Except for the 6.6 m membrane, the experimentally determined pcap values are higher than the theoretical values cal- culated from the Laplace equation. Under given conditions the emulsions with a very narrow droplet size distribution were prepared over a wide range of mean droplet sizes (Fig. 4). The span of the droplet size distribution of 0.260.45 (Table 2) was much lower than that reported for ceramic membranes. As an example, Joscelyne and Trgrdh 11 obtained the span of 0.891.6 using ceramic membranes with a mean pore size of 0.1 Droplet diameter d / m 110100 10-2 10-1 1 0 20 40 60 80 100 Cumulative volume Q 3 / vol % Fig. 4. Droplet size distribution curves at ptm 1.1pcap: , mean pore size dp = 6.6 m; , dp = 5.0 m; , dp = 2.5 m; , dp = 1.4 m; , dp = 0.4 m. Volume frequency q3 / m-1Cumulative volume Q3/vol% G.T. Vladisavljevic, H. Schubert / Desalination 144 (2002) 167172171 Pore size, dp, m0.41.4 2.5 5.0 6.6 pcap, kPa 185 48 24 10 5 d3,2, m 1.44.6 8.5 14.7 23.9 Span 0.450.300.440.37 0.26 Jd, l m2 h1) 0.72.3 2.9 4.3 6.6 Table 2 The experimentally found capillary pressures, pcap, and emulsification results at ptm 1.1pcap and = 12 vol% for different mean pore sizes 0.5 m. However, the dispersed phase flux in their experiments was 15270 kgm2h1. Williams et al. 12 obtained the span of 0.83 at the oil flux of 8 l m2h1 using ceramic membrane with a mean pore size of 0.5 m. The mean droplet size was found to be 3.5 times larger than the mean pore size, if ptm 1.1pcap (Fig. 5). The droplet/pore diameter ratio of 3.5 is similar to 3.25 reported by Nakashima et al. 1 for SPG membrane. Using Eq. (3), it was found that droplets were formed simultaneously at only 1.6 2.6% of the pores (Table 3). Taking a mean value of 0.58 and k = 0.02, one obtains from Eq (5): ddrop/dp 8. Thus, the condition for unhindered droplet growth is satisfied in the given case. The average droplet formation times of 0.61.8 s (Table 3) were similar to 11.5 s found by Schrder and Schubert 8 for Tween 20 and a ceramic membrane. The longest droplet formation time for the 6.6 m membrane is a consequence of largest droplet volume in that case. The mean droplet size was almost independent of the dispersed phase content up to 20 vol% (Fig. 6), although the dispersed phase flux through the membrane increased with time. This trend was also found by Katoh et al. 13 in the preparation of food emulsions using SPG membranes. The small decrease of the mean droplet size in the experiments with the 6.6 m membrane can be explained by the disruption of droplets during recirculation. At the smaller cross-flow velocity of the continuous phase (0.9 ms1), the disruption was less pronounced but the mean droplet size was higher. Table 3 Calculations of the fraction of active pores, k and the average droplet formation time, tf at ptm 1.1pcap and = 12 vol % for different mean pore sizes Pore size, dp, m 0.4 1.4 2.5 5.0 6.6 k 0.019 0.020 0.016 0.020 0.026 tf , s 0.6 0.7 0.8 0.9 1.8 01234567 0 4 8 12 16 20 24 Mean pore size dp / m d3,2 = 3.5dp Fig. 5. Relationship between mean pore size and mean droplet size at ptm 1.1pcap. 05101520 10 14 18 22 26 30 Dispersed phase concentration / vol % 2.5 m-Membrane 5 m-Membrane 6.6 m-Membrane, 1.4 m/s 6.6 m-Membrane, 0.9 m/s Fig. 6. The influence of dispersed phase content on mean droplet size at ptm 1.1pcap for different mean pore sizes. 172G.T. Vladisavljevic, H. Schubert / Desalination 144 (2002) 167172 5. Conclusions The SPG membrane can be successfully used to produce O/W emulsions with a very narrow droplet size distribution, on the condition that the transmembrane pressure is not much higher than the capillary pressure. Under these conditions the mean droplet size is 3.5 times larger than the mean pore size and not significantly affected by the dispersed phase content up to 20 vol%. However, the transmembrane flux is very low, because at any moment only about 2% of the pores are active. Acknowledgement This work was financially supported by the Alexander von Humboldt Foundation, Bonn, Germany. Symbols dp Mean pore size of the membrane, m ddrop Droplet diameter, m d3,2 Mean Sauter diameter, m d4,3 Volume-weighted mean droplet diameter, m Jd Dispersed phase flux through the mem- brane, ms1 Jw Pure water flux through the membrane, ms1 k Fraction of active pores lp Pore length, m pcap Cappilary pressure, Pa Rm Hydraulic membrane resistance, m1 tf Average droplet formation time, s vw Mean water velocity in the pores, ms1 z Distance between the centres of adjacent active pores, m Greek ptm Transmembrane pressure, Pa m Membrane thickness, m Membrane porosity d Viscosity of dispersed phase, Pas w Viscosity of water, Pas Mean tortuosity factor of pores Dispersed phase concentration in emul- sion, vol % References 1T. Nakashima, M. Shimizu and M. Kukizaki, Mem- brane Emulsification, Operation Manual, Industrial Research Institute of Miyazaki Prefecture, Miyazaki, Japan, 1991. 2H. Karbstein and H. Schubert, Developments in the continuous mechanical production of oil-in-water macro-emulsions, Chem. Eng. Process., 34 (1995) 205211. 3T. Nakashima and M. Shimizu